Date:         Fri, 21 Jul 2006 17:21:16 +0000
Reply-To:     toby dunn <tobydunn@HOTMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         toby dunn <tobydunn@HOTMAIL.COM>
Subject:      Re: Proc Freq Question
Comments: To: scottwh@AMAZON.COM
In-Reply-To:  <1934F70034988F4FB907B7D364F8D412010CA6F3@exchg-sea5-04.ant.amazon.com>
Content-Type: text/plain; format=flowed

Scott ,

Looks to me like proc tabulate would be a better fit than proc freq.



Toby Dunn

'They say that Spanish is the Language of Love,
I Loved the way it rolled off of her tongue'

'I dont know what she said but I loved the way she said it'






From: "White, Scott" <scottwh@AMAZON.COM>
Reply-To: "White, Scott" <scottwh@AMAZON.COM>
To: SAS-L@LISTSERV.UGA.EDU
Subject: Proc Freq Question
Date: Fri, 21 Jul 2006 10:16:17 -0700

Hi Folks,

Let's say I have a data set with just three variables: observation id (oid),
binary predicted value (pred) and dollar amount of transaction which we are
making a prediction for (amt). Is there a way with just one proc statement
to output a table that shows for both types of predictions both how many
orders and what the total dollar amount was for each type? Basically the
format of the desired output looks like this:

Target|# Orders|$ Amount
Yes|35|$1000
No|240|$10,000

The way I do it right now is to use two proc freqs one that uses the
statement "weight amt;" and one that doesn't and then I merge the two to get
the desired result.